Title
Predicting Efficiency of Commercial Banks in Vietnam: A DEA and Machine Learning Approach
Authors
Abstract
Purpose: This study investigates the effectiveness of a hybrid Data Envelopment Analysis (DEA) and Machine Learning (ML) approach in predicting the efficiency of commercial banks in Vietnam.
Methodology: A two-stage model is proposed. First, DEA is employed to evaluate bank efficiency from 2012 to 2021 using data from annual reports. Second, various ML algorithms (ANN-MLP, linear regression, random forest) are used to forecast efficiency scores based on the DEA results. The performance of each ML model is compared to identify the most effective approach.
Findings: The findings suggest that the ANN-MLP model outperforms other ML methods in predicting bank efficiency.
Research limitations/implications: The study utilizes data from a specific timeframe and may be limited by potential inaccuracies in financial statements. Future research could extend the time period and explore additional data sources.
Practical implications: The proposed DEA-ML model can be a valuable tool for bank managers and policymakers to assess and predict bank efficiency, ultimately leading to improved decision-making, greater efficiency, and enhanced competitiveness. The findings might be generalizable to other bank types in similar contexts.
Social implications: This research contributes to the development of DEA-ML models, potentially influencing practices in bank efficiency measurement and leading to a more robust financial system.
Originality/value: This study offers a novel approach for combining DEA and ML techniques to predict commercial bank efficiency in Vietnam. The findings demonstrate the potential of ANN-MLP for financial applications and provide valuable insights for bank management and financial regulation.
Keywords
Vietnamese commercial banks; Efficiency; Data Envelopment Analysis; Machine learning.
Classification-JEL
G21, C14, C53